New technology simulates adverse weather

PhD researcher Chenghao Qian has developed WeatherEdit, which builds on real-world simulations to help researchers understand the effects of harsh weather conditions.

The application can add rain, snow and fog to 3D simulations to support vehicle testing and training in conditions that are otherwise rare or dangerous to collect.

Climate change is making the weather more unpredictable. This creates serious challenges for technologies like autonomous vehicles and robotics.

Sudden fog, heavy rain, or snow can cause them to behave unpredictably or even fail.

Yet collecting real-world data in such conditions is difficult, time-consuming, and unsafe.

That’s why Chenghao Qian and his colleagues in the Human Factors & Safety Group developed WeatherEdit.

Chenghao said: “As climate change leads to more extreme weather, evaluating intelligent autonomous systems under such conditions is important.

“Instead of waiting for extreme weather to test safety, we can now create it on demand. WeatherEdit makes intelligent systems smarter—and safer—under real-world challenges.”

Read the paper titled "WeatherEdit: Controllable Weather Editing with 4D Gaussian Field," co-authored by Gustav Markkula, Wenjing Li, and Yuhu Guo.